Linear Discriminant Functions in Connection with the micro-RNA Diagnosis of Colon Cancer

Early detection (localized stage) of colon cancer is associated with a five-year survival rate of 91%. Only 39% of colon cancers, however, are diagnosed at that early stage. Early and accurate diagnosis, therefore, constitutes a critical need and a decisive factor in the clinical treatment of colon cancer and its success. In this study, using supervised linear discriminant analysis, we have developed three diagnostic biomarker models that—based on global micro-RNA expression analysis of colonic tissue collected during surgery—can discriminate with a perfect accuracy between subjects with colon cancer (stages II–IV) and normal healthy subjects. We developed our three diagnostic biomarker models with 57 subjects [40 with colon cancer (stages II–IV) and 17 normal], and we validated them with 39 unknown (new and different) subjects [28 with colon cancer (stages II–IV) and 11 normal]. For all three diagnostic models, both the overall sensitivity and specificity were 100%. The nine most significant micro-RNAs identified, which comprise the input variables to the three linear discriminant functions, are associated with genes that regulate oncogenesis, and they play a paramount role in the development of colon cancer, as evidenced in the tumor tissue itself. This could have a significant impact in the fight against this disease, in that it may lead to the development of an early serum or blood diagnostic test based on the detection of those nine key micro-RNAs.